ChatGPT referral traffic jumps 25x, but news sites are still bleeding

For years, news sites lived and died by Google search clicks. But when an AI directly gives you the gist of the morning headlines, clicking through to a publisher’s website feels like extra work nobody wants to do.

According to a fresh report from intelligence firm Similarweb, the bill has finally arrived for digital media. While ChatGPT is sending way more traffic to news outlets than it used to, that boost doesn't come close to plugging the massive hole left behind by search engines.

A 25x surge that still falls short

Ever since Google rolled out AI Overviews in May 2024, web news searches that end without a single click to a publisher jumped from 56% to nearly 69% as of May 2025.

That dynamic drove overall organic search traffic off a cliff. Publishers saw visits plunge from a peak of over 2.3 billion in mid-2024 to under 1.7 billion today.

At the same time, users are turning directly to chatbots for information. News-related prompts inside ChatGPT surged 212% between January 2024 and May 2025. App users more than doubled over the past six months, while web visitors jumped 52%.

That shift did produce a silver lining in raw numbers. ChatGPT referrals to news sites soared from under 1 million in the first five months of 2024 to more than 25 million in the same period of 2025. That is a massive 25x jump.

Yet when an industry loses over 600 million organic search visits, picking up an extra 24 million AI clicks is hardly a rescue.

The winners, the losers, and the lawsuits

ChatGPT isn't sharing the love equally across the board.

Outlets like Reuters saw ChatGPT referral traffic climb 8.9% year-over-year. The NY Post gained 7.1%, and Business Insider went up 6.5%.

Then there is The New York Times. The paper, which is actively suing OpenAI for scraping its stories without consent, saw just a 3.1% bump in ChatGPT referrals. While still clinging to the top ten, its growth trails competitors by a wide margin.

Topics like stocks, finance, and sports dominate news prompts, but queries around politics, economics, and weather are climbing fast.

To survive the bleeding, outlets are scrambling. Google recently pushed out Offerwall, letting publishers test micropayments or newsletter signups instead of traditional ads. Others are resorting to tighter paywalls, brutal rounds of layoffs, or closing up shop altogether. OpenAI CEO Sam Altman recently admitted on The NYT’s Hard Fork podcast that entire job categories could vanish, noting there will be real pain.

Why it matters

Publishers can no longer afford to view AI referrals as a replacement for SEO. If your business model relied on open web search traffic, that engine is breaking down fast. The priority has to pivot toward direct reader relationships, native apps, and proprietary newsletters that AI summaries cannot intercept.

For the tech industry, this dynamic highlights a looming parasite problem. Large language models feed on deep reporting and breaking coverage. By answering queries directly and slashing publisher visits by hundreds of millions, AI companies risk starving the very content creators they rely on for fresh data.

We think tools like Google's Offerwall look like modest band-aids on a severed artery. Big tech keeps users trapped on summary pages while telling publishers to squeeze whatever micropayments they can from the remaining trickle. Unless platforms establish sustainable revenue-sharing mechanisms, the decline of original journalism will eventually degrade the quality of AI answers themselves.